Case Study: Cutting Bright Beer Tank CIP Water and Caustic With Rotary Spray Heads
A regional brewery cut bright beer tank CIP water and caustic by swapping static spray balls for rotary spray jet heads. Sizing logic you can reuse.
A bright beer tank holds finished beer, so the wall behind the next fill has to be cleaner than any upstream vessel. The tank also runs under a CO2 blanket, so every cleaning device works against backpressure. A regional craft brewery in North America cleaned three 20 kL bright beer tanks with static spray balls. Each cycle ran a 20 minute caustic wash and passed about 3,000 L of wash liquor through the tank. The wash passed most swabs and failed on the same shadow behind the racking arm.
That pattern is common in brewery tank cleaning, and it costs more than the water it wastes. Published brewing references put cleaning and sanitising at a large share of plant water use. This case covers one change at that plant, along with the sizing work that kept the new heads turning. Every figure is a working figure for this class of pressurized vessel.
Case Snapshot
The table below is the generic shape of the installation, not a plant record. Each value is a realistic working figure for a pressurized bright beer tank.
| Item | Value |
|---|---|
| Industry | Beverage, regional craft brewing |
| Duty | CIP of pressurized bright beer tanks between fills |
| Vessels | Three vertical tanks, 20 kL each, in the 10 to 30 kL class |
| Wetted surface | About 130 m² per tank, shell, heads, and internal fittings |
| Headspace | CO2 blanket held at 0.8 to 1.5 bar through the cycle |
| Original setup | Three static spray balls, about 150 L/min at 1.5 to 2 bar |
| Cleaning head chosen | Two fluid-driven rotary spray jet devices, about 40 L/min at 3 bar |
| Caustic wash | 2 percent w/w sodium hydroxide at 60 to 80 °C |
| Acid wash | Nitric and phosphoric blend for beer stone |
| Sanitiser | Cold PACS at 150 to 300 ppm, 10 to 20 minute contact |
| Compliance | Food-contact duty, documented CIP, allergen control |
Wetted surface drives coverage. About 130 m² of shell, heads, and internal fittings is the coverage target for this class, and every flow calculation below starts from it. A bright beer tank also carries a racking arm, a carbonation stone, a sample valve, and a CIP standpipe. Each fitting casts a shadow that a static ball cannot sweep.
Pressurization is the second defining feature. The tank holds 0.8 to 1.5 bar of CO2 through much of its working life, and many breweries clean it without venting. Published nozzle data shows throw falling as gas density rises, so a head rated in air loses reach inside a pressurized tank.
The Challenge
The residue on a bright beer tank differs from the residue on a fermenter. Beer arrives here filtered and cold, so the soil is a thin film of protein, hop acid, and calcium oxalate. That film dries hard on the upper shell at the fill line. Brewers call the mineral layer beer stone, and caustic alone will not dissolve it.
The shadow problem decided the outcome. A static spray ball throws liquid in every direction at once. Liquid that reaches the far side of the racking arm has crossed a longer path, so it arrives with less energy. The wall stays wet and the swab still fails. In brewery tank cleaning, that is the classic under-cleaning signature.
The consequence is worse here than on any other vessel in the brewery. A failed clean puts a 20 kL batch of finished beer at risk. Twenty thousand litres is roughly 60,000 bottles that have already absorbed every upstream cost. The plant could not enter the tank to fix it by hand either. A vessel under a CO2 blanket is an asphyxiation hazard, so entry needs a purge, a permit, and two people on standby. In practice the only correction was a second full automated cycle.
The cost picture was the second pressure. The old caustic phase ran 20 minutes at about 150 L/min, which is 3,000 L of wash liquor per tank per cycle. At 2 percent w/w caustic that is about 60 kg of sodium hydroxide sent to the drain. Published brewing benchmarks put total water use at 3 to 6 L per litre of beer, and cleaning takes a large share of it. The chemical bill tracked wash volume almost directly.
Documentation closed the case for change. The plant could not prove the automated cycle reached the shadow behind the racking arm. The audit question never changes. What were the flow, the rotation, the temperature, and the time on this cycle? Operator judgement is not an answer, and a re-clean is the only correction left once a tank is signed off.
Why a Static Spray Ball Wets a Tank Better Than It Cleans It
Published reference data puts static spray ball duty at roughly 130 to 200 L/min at 1.5 to 2 bar. The ball spreads that flow over a full sphere, so the wall receives a thin sheet with little momentum behind it. Coverage is excellent and impact is minimal. Published wetting targets for cleaning coverage run about 0.4 to 0.6 L/min per m² of surface. The static bank cleared that band with room to spare.
The arithmetic makes the point. At 150 L/min over 130 m², the static bank delivers about 1.15 L/min per m². That is more than twice the top of the published wetting band. The tank still failed swabs. Wetting is a coverage measure, and coverage was never the problem. Removing a dried film needs repeated impact at high local velocity, and that is what the ball cannot deliver.
Impact is not pressure, and it is not flow either. The kinetic energy of a jet rises with the square of its velocity. Velocity rises when the same flow passes through a smaller orifice. At 10 L/min through a 2 mm orifice the liquid leaves at about 53 m/s. The same flow through a 4 mm orifice leaves at about 13 m/s. Energy per litre differs by roughly sixteen times.
Distance compounds the loss. Flow from a static ball spreads over a sphere, so intensity falls with the square of distance from the nozzle. Published hygiene practice expects several impacts at every surface point per rotation period, with at least four impacts as the commonly cited figure. A ball completes no rotation and lays down no pattern, so there is nothing to count.
The rotary jet head guide sets out the pattern logic for this device class in more detail.
What a Rotary Spray Jet Device Changes in a Pressurized Tank
Rotary spray jet devices sit in a different published envelope. The class runs roughly 25 to 60 L/min at 2 to 4 bar. That is a fraction of the flow a static bank needs for the same vessel. The device turns the smaller flow into two or four coherent jets and sweeps them through a fixed pattern. Each point on the wall is hit many times per rotation instead of being flooded once.
The wetting test is read differently for this device class. Across the whole 130 m² the head averages about 0.31 L/min per m², below the band published for spray-ball coverage. That average is not the sizing test for a jet device. Inside the swept band the impinged strip sees a wetting rate many times the target. The strip walks the whole interior inside one rotation period. Coverage then satisfies the impact rule, with at least four passes at every point.
Rotation is what makes the cycle repeatable. A fluid-driven head has no motor, no wiring, and no gearbox inside the tank. The cleaning liquid spins the rotor, and the internal gear set fixes the pattern. Rotation speed follows flow, which suits CIP because the loop pump runs steady. A pattern that repeats exactly is a pattern that can be logged.
The pressurized headspace changes the device behaviour. A jet inside a dense gas loses energy faster than the same jet in air, and published throw figures are usually quoted in air. The plant therefore sized the pattern against working headspace pressure rather than the catalogue figure. The head enters through a top port and aims down the shell, because no safe way exists to reach inside a gassed tank.
The Solution
The change was small in hardware terms and large in effect. Three static spray balls came out of each tank. Two rotary spray jet devices went in, mounted on the CIP standpipe in the tank top. The pair runs at about 40 L/min in total at 3 bar at the head. That sits inside the published 25 to 60 L/min and 2 to 4 bar envelope for the class. The loop pump already on site was re-rated rather than replaced.
The cycle kept the phases the plant already used. A short pre-rinse lifts the loose film. A caustic wash at 2 percent w/w sodium hydroxide and 60 to 80 °C breaks down protein and hop residue. An intermediate rinse clears the caustic. An acid wash at a nitric and phosphoric blend removes beer stone. A cold PACS sanitiser at 150 to 300 ppm and 10 to 20 minutes of contact follows the final rinse. Published brewery tank cleaning guidance puts caustic at 1 to 3 percent w/w and 60 to 80 °C.
The caustic contact time halved. The old cycle ran 20 minutes to give the ball time to soak the film loose. The new cycle runs 10 minutes, because the patterned jets remove the film in a few passes. That single change carries most of the water and chemical saving in this case. It is also the part that depends on impact rather than volume.
The acid step matters more here than on most CIP duties. Beer stone is calcium oxalate, and caustic does not dissolve it. A tank that runs only caustic and sanitiser builds a mineral layer that grows with every cycle. The acid phase at 3 minutes through the same heads keeps the upper shell clear at the fill line.
Material followed the duty. Wetted parts are 316L stainless with an Ra ≤ 0.8 µm finish on product contact surfaces. This brewery stayed on 316L because its chemistry is standard caustic, acid, and peracetic sanitiser. Every dead leg was removed from the standpipe assembly, because a device that cannot drain becomes a contamination source.
The retrofit did not change the floor plan. The heads sit inside the tank, so no bay was rebuilt and no extra pump was added. The CIP tank washing systems guide covers the same loop architecture for CIP tank cleaning at this scale.
Head Layout and Flow Budget on a Bright Beer Tank
Placement follows the same rule as any vessel. Mount the heads about one third of the liquid height above the bottom, aimed at the worst-soiled zone. On this duty that zone is the upper shell at the fill line. Keep the heads above the liquid level so the jets travel through gas. A jet that travels through liquid loses most of its energy within the first metre.
Two heads beat one on a wide bright beer tank. A single head must reach both the top head and the bottom knuckle, and that span needs a longer throw. A pair mounted off the tank axis splits the span and covers the shell in one rotation. The racking arm and the carbonation stone still cast shadows, and the off-axis mounting is what sweeps them.
The flow budget closes the sizing. Published wetting guidance and the rotary device envelope set the flow, and the phase times set the volume. At 40 L/min for 10 minutes the caustic phase puts 400 L through the tank. At the old 150 L/min for 20 minutes the same phase put 3,000 L through it. That ratio is the whole water argument, and a plant can check both numbers on its own pump curve.
The full cycle balance for one tank looks like this.
| Phase | Static spray balls | Rotary spray jet devices | Why it moves |
|---|---|---|---|
| Pre-rinse | 3 min at 150 L/min, 450 L | 2 min at 40 L/min, 80 L | A pattern reaches the wall sooner than a flood |
| Caustic, 2 percent at 60 to 80 °C | 20 min at 150 L/min, 3,000 L | 10 min at 40 L/min, 400 L | Impact strips the film in fewer passes |
| Intermediate rinse | 3 min at 150 L/min, 450 L | 2 min at 40 L/min, 80 L | Less volume to clear the caustic |
| Acid for beer stone | 5 min at 150 L/min, 750 L | 3 min at 40 L/min, 120 L | Same pattern, lower volume |
| Cold PACS sanitiser | 15 min at 150 L/min, 2,250 L | 12 min at 40 L/min, 480 L | Contact time is set by the sanitiser, not the flow |
| Cycle total | About 6,900 L | About 1,160 L | Flow ratio times phase time |
Those totals assume both setups run the caustic and acid phases one-shot. A plant that recirculates its caustic charge would not spend 3,000 L of fresh caustic per cycle. Its freshwater saving would be smaller than the table suggests. The chemical saving then tracks the number of charges per shift rather than the volume through the tank. This plant ran a one-shot caustic, which is why the volume figure is also the chemical figure.
Why Undersized Pumps Stall Rotary Heads
A fluid-driven rotary head has a minimum flow. The internal rotor converts flow into rotation. Below the published minimum the head turns slowly, then stalls against the friction of its own seal. A stalled head does not stop the cycle. It lays a partial pattern and the controller logs a normal phase time, so the tank fails a swab three days later. The commonest cause of a failed rotary retrofit is a pump that cannot hold rated flow at the head pressure.
Flow through a nozzle rises with the square root of the pressure drop. Doubling the pressure therefore raises flow by about 41 percent. A head rated at 25 to 60 L/min and 2 to 4 bar needs a pump that holds that point. A loop that will not reach 40 L/min at 3 bar at the head will not keep this head turning.
Backpressure is the hidden term in that calculation. Nozzle pressure equals pump discharge minus line loss minus headspace pressure. A tank held at 1 bar of CO2 removes 1 bar from the nozzle before any friction is counted. The same pump that delivers 3 bar in an open vessel delivers 2 bar in a pressurized one.
Throw derates with gas density as well. Published nozzle data shows spray reach falling as the surrounding gas becomes denser. A bright beer tank under CO2 is denser than the air behind the catalogue figures. Test the pattern at working headspace pressure, and keep margin in the flow budget. Brewery tank cleaning loops fail on flow more often than on pressure.
Return velocity matters more than headline pressure. Published CIP practice sets return line velocity around 1.5 to 2 m/s. The liquid then runs as a moving film that carries soil back to the tank. A loop that meets a pressure target but starves on return velocity leaves soil in the low runs and the bottom knuckle. Check both numbers after the retrofit and log them with the phase time.
Confirm the rotation at first start with a visual or acoustic check. Spray nozzles wear before they fail, and published wear data shows flow rising 5 to 10 percent at end of life. A worn nozzle changes the pattern, so a periodic flow check belongs on the maintenance schedule.
The Results
No site audit figures appear below. These are working figures for this class of vessel, checked against published ranges for rotary spray jet devices replacing static spray balls.
| Metric | Static spray balls | Rotary spray jet devices | Basis |
|---|---|---|---|
| Wash liquor per tank cycle | About 6,900 L | About 1,160 L | Flow at the head times phase time |
| Caustic phase volume | About 3,000 L | About 400 L | 150 L/min for 20 min, then 40 L/min for 10 min |
| Caustic charge at 2 percent | About 60 kg | About 8 kg | Dosed on wash volume on a one-shot phase |
| Caustic contact time | 20 min | 10 min | Patterned impact strips the film in fewer passes |
| Water per litre of beer | Higher share of plant use | Lower share | Published benchmarks put brewing water at 3 to 6 L per litre |
| Impacts per surface point | No repeating pattern to count | At least four per rotation | Published hygiene practice for jet devices |
| Added pump | None | None, loop re-rated | Fluid-driven heads use the process flow only |
| Manual entry | Needs a purge and a permit | Not required | A CO2 blanket bars entry while the tank is gassed |
The table is a range picture, not a promise for a specific brewery. The result depends on the residue, the vessel geometry, the pump curve, and the sanitiser contact time. What the change buys is repeatability, because the pattern is fixed by the head geometry.
The acceptance test the plant used is the one to copy. Log flow, rotation, temperature, and time for each phase. Compare the new cycle against two weeks of baseline data from the old one. Swab the fill line, the top head underside, the racking arm, the carbonation stone, and the bottom knuckle.
Reading the Numbers Against Published Ranges
Every figure above sits inside a published band.
The caustic wash runs at 1 to 3 percent w/w and 60 to 80 °C in published brewery CIP practice. This plant ran 2 percent. The cold PACS sanitiser runs at 150 to 300 ppm with 10 to 20 minutes of contact, and the cycle sat inside that window. Rotary spray jet devices in this class are published at 25 to 60 L/min and 2 to 4 bar. The chosen point of 40 L/min at 3 bar sits in the middle. Static spray ball duty is published at 130 to 200 L/min at 1.5 to 2 bar. The old bank sat inside that band as well.
The wetting figure needs care. Published guidance puts the wetting target at 0.4 to 0.6 L/min per m² for cleaning coverage. Average flow over the whole vessel is the wrong test for a jet device. The relevant tests are the wetting rate inside the swept band and the impact count at each point. Both pass, with at least four impacts per rotation. That is why the head meets the published coverage requirement at a fifth of the static flow.
Two caveats protect the numbers. First, the volume comparison assumes a one-shot caustic phase. Recirculation changes that, so compare fresh water consumption rather than total wash volume. Second, the flow ratio only holds if the pump keeps the head turning at working headspace pressure. Measure flow at the head with the tank at its normal CO2 pressure before the change is made permanent.
This is the pattern to copy in any brewery tank cleaning trial, so run the comparison against a logged baseline. Two weeks of the old cycle gives a water figure, a caustic figure, a cycle time, and a swab pass rate. The new cycle either beats that baseline on the automated sequence or it does not.
Why This Matters
The logic here is not specific to beer. Any tank that holds a finished product runs under pressure and cannot be entered for manual cleaning. Coverage, impact, and proof then all rest on the installed head. Start with the wetted area and the residue. Pick the head class next, then check the pump against working pressure at the nozzle. That order rules out most of the catalogue before flow is discussed.
Three numbers carry the case. About 130 m² of wetted surface sets the coverage target. A published wetting band of 0.4 to 0.6 L/min per m² sets the flow floor for spray-ball coverage. At least four impacts per rotation sets the floor for jet devices. A headspace at 0.8 to 1.5 bar of CO2 sets the derating on throw and the pressure the pump has to beat.
For brewery tank cleaning at this scale, the change pays back through the cycle rather than through the hardware. A fluid-driven head uses the process flow, so the rotary spray jet pair costs less than the pump upgrade some plants expect. The saving lands in three places. Less wash liquor per cycle, less caustic per cycle, and a cycle that fits the production window. A 10 minute caustic phase against a 20 minute phase is production time recovered on every tank turn.
The hardware for this duty sits in the tank cleaning products range. The sizing path above is the same one that produced this result. Send the vessel drawing, the wetted area, the residue, and the cleaning chemistry. Add the pump curve with the flow at pressure at the connection through the contact page. BoreJet comes back with the head count, the flow point, and the cycle that fits the tank.
Your Duty May Differ
Send the Vessel Drawing and the Pump Curve.
We size the head class, the count and the operating point for your tank, not a catalogue guess. A coverage test on the actual vessel beats any estimate.